Van Waerebeek: Variation in skull morphology of Lagenorhynchus obscurus 



761 



Southwest African versus New Zealand popu- 

 lation Results of two-sided /-tests comparing adult 

 skulls from southwest Africa and New Zealand are pre- 

 sented in Table 6. Skulls from these regions do not 

 differ in size, as indicated by condylobasal length (/=0.48, 

 P=0.63) and zygomatic width i/=-0.074, P=0.94). Nev- 

 ertheless, the two populations are strikingly divergent 

 in dentition. Mean tooth width for New Zealand dusky 

 dolphins (3.56 mm, SD=0.35, N=32) is highly signifi- 

 cantly smaller (/=-7.14, P<0.0001) than in SW African 



dusky dolphins (4.18 mm, SD=0.37, N=36); the 

 difference is visually obvious even from a quick 

 inspection of skulls. In addition, New Zealand 

 crania have on average 1.5-1.8 teeth more in 

 each row than African skulls (two-sided /-tests, 

 P<0.001). Maximum tooth counts in the four 

 rows range 32-34 (SA) and 36-39 (NZ). Finally, 

 SW African crania have a significantly lower 

 supraoccipital crest (/=3.43, P=0.005), slightly 

 wider external nares and temporal fossa and, 

 possibly, a somewhat higher braincase 

 (0.05<P<0.01) than the New Zealand specimens. 



Peruvian versus Chilean population Dusky 

 dolphin crania from central Peru and north- 

 ern Chile showed no meaningful overall size 

 differences, as indicated by CBL and ZYG 

 (Table 6). The antorbital process (APL) and 

 the lower tooth row (LTRL) however were very 

 significantly shorter (/-test, P<0.005) and mean 

 tooth counts were lower in Chilean animals. 

 Small but statistically discernible deviations 

 (P<0.05) were detected also in PROW, PTEL, 

 and RAL. These data suggest that at least 

 some genetic distance exists between L. 

 obscurus stocks from central Peru and north- 

 ern Chile, but it is recommended that the 

 sample size for Chile be increased before defi- 

 nite conclusions be made. 



Multivariate analysis Results of the dis- 

 criminant analyses (DISCRIM) with a maxi- 

 mum of characters (21) fully confirm findings 

 from bivariate techniques (Table 7); scores 

 on first and second axis are shown in Figure 

 3. The first discriminant function differenti- 

 ated well between population pairs Peru/ 

 Chile and New Zealand/SW Africa; the sec- 

 ond function largely separated SW Africa and 

 New Zealand stocks (Fig. 4), whereas the 

 third function projected the Chile group at 

 the positive extreme of the z-axis (see group 

 centroids, Table 7). The discriminant func- 

 tions successfully predicted the actual group 

 most of the time, i.e., from 85.7% of cases in 

 Chile to 91.7% in New Zealand. When Chilean and 

 Peruvian animals were combined into a single 

 "Southeast Pacific stock," 96.5% ( 137 of 142) of these 

 were correctly allocated. Only two of 24 (8.3%) New 

 Zealand skulls were misclassified as SW African but 

 they were never mistaken for SE Pacific animals. 

 Two skulls out of 22 (9.2%) from SW Africa were 

 erroneously assigned, one each to the New Zealand 

 and Peru stocks. The highest percentages of 

 misclassifications ( 10.4% and 14.3%) occurred among 



